Unverified50% confidenceFactExact time
In QSA, a C4 block refers to every 4 adjacent tokens being organized into a single computational unit, with attention selection performed at the block level rather than the token level.
1
Sources
50%
Confidence
Long-term
Relevance
9/15/2026
First Seen
Sources
Related Entities
Related Claims
Unverified自注意力通过将输入序列映射为查询Q、键K、值V三组向量，计算Q与K的点积相似度后经Softmax归一化得到注意力权重，再加权汇聚V向量72% similarVerifiedGrouped Query Attention (GQA) and Multi-Query Attention (MQA) reduce KV cache memory usage during inference by having multiple query heads share the same group of key/value heads, and are standard optimizations in models such as Llama 3 and Mistral.68% similarUnverifiedGGUF格式中的Q4_K_M采用分组量化（Group Quantization）策略，每组独立计算缩放因子和零点65% similarUnverifiedThe concept of KV Cache groups originates from techniques such as GQA (Grouped Query Attention), where multiple query heads share the same set of KV heads.64% similarUnverifiedCurrent mainstream 4-bit quantization schemes (such as GPTQ and AWQ) universally adopt a group quantization strategy with group size=128.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/923184API
curl https://kongchang.com/api/v1/knowledge/claims/923184MCP
get_claim(id=923184)